FCT: DEE - Dissertações de Mestrado
URI permanente para esta coleção:
Navegar
Entradas recentes
- Estudo do Enquadramento Legislativo e Identificação de Barreiras na Criação de Comunidades de Energia Renovável em PortugalPublication . Gabriel, João Pedro Palaio de Almeida; Amaro, NunoO aquecimento global, as consequentes mudanças climáticas e a crise energética devido à depen- dência de combustíveis fosseis para a produção de energia são uns dos maiores problemas da atualidade, mas estes podem ser mitigados através de uma transição para energia produzida a partir de fontes reno- váveis. As comunidades de energia são apresentadas como uma ferramenta importante para ajudar nesta transição energética, promovendo a produção de energia renovável descentralizada enquanto ao mesmo tempo beneficiando as correspondentes comunidades de uma forma económica, social, e ambiental, possibilitando o cidadão a ter um papel ativo na transição energética. As comunidades de energia foram oficialmente reconhecidas como Comunidades de Energia Renovável (CER) na Diretiva das Energias Renováveis, ou RED II, que obriga cada Estados-Membros (EM) da União Europeia (UE) a transpor para as suas correspondentes legislações estas diretivas, habilitando assim a implementação de CER em todos os EM. A implementação destas CER é, atualmente, aquém do ideal, sendo esta implementação atrasada devido a barreiras tanto a nível regulatório como a nível técnico, económico e social. A presente dissertação tem como objetivo a identificação das possíveis barreiras à implementação de CER, analisando o estado atual da transposição das Diretivas Europeias relativas às CER, do enqua- dramento legislativo e quadro facilitador da implementação destas, tanto de Portugal como de outros Estados-Membros da União Europeia. Após a identificação destas barreiras, são apresentadas possíveis medidas mitigadoras para cada vertente de barreira. Em Portugal, as principais barreiras que afetam a criação de CER são de vertente regulatório, que inclui o processo complexo e moroso de obtenção das licenças necessárias, tendo sido identificada como medida de mitigação a digitalização e automatização parcial do processo de licenciamento. Foram também identificadas barreiras e medidas de mitigação a nível técnico, económico e social.
- Políticas de eficiência energética para edifícios na Europa e no Brasil - Proposta de lei brasileira para a eficiência de edifícios e plataforma de suportePublication . Machado, Marcus Antonio; Gonçalves, Maria Margarida; Faria, PaulinaA presente dissertação analisa e compara as políticas de eficiência energética em edifícios na União Europeia (UE) e no Brasil, destacando estratégias regulatórias, tecnológicas e económicas que visam a redução do consumo energético e das emissões de gases com efeito de estufa. O trabalho identifica as principais convergências e divergências entre os dois contextos, com enfoque em mecanismos legislativos, instrumentos de financiamento e critérios de sustentabilidade. Concluiu-se que, embora o Brasil disponha de instrumentos relevantes, o enquadramento europeu apresenta maior coerência normativa e integração entre políticas públicas, evidenciando a necessidade de um modelo brasileiro mais estruturado, especialmente no que respeita à reabilitação energética do parque edificado. Neste contexto, é proposto um anteprojeto de lei federal para o Brasil, inspirado em experiências europeias e adaptado às especificidades climáticas, económicas e sociais do país. Complementarmente, foi desenvolvida uma plataforma digital de apoio à divulgação de informação e boas práticas, contribuindo para o debate académico, para disseminação de informação na sociedade e para a promoção da sustentabilidade no setor da construção civil.
- NOVAMOB: An Open-Source Simulation Framework for Mobile Robot Navigation and ValidationPublication . Carvalho, Rafael Santos; Peres, RicardoIn this day and age, Robotics is gaining increasing significance in various sectors, such as the manufacturing, automotive and agriculture industries, due to its ability to efficiently, accurately, and safely perform tasks previously assigned to humans. Consequently, the demand for experienced professionals in this sector is also increasing, making it essential to provide students in the field with practical experience. However, providing future professionals with practical, hands-on experience presents several challenges. Traditional training methods with physical robots involve high costs and, at times, considerable complexity, leading many educational institutions to struggle to provide their students with the necessary practical experience, therefore increasing the barrier between theoretical knowledge and real-world application. With this in mind, the NOVAMOB initiative aims to democratize access to robotics education by developing modular, open-source robots at a more accessible cost, thus facilitating the study of Robotics in universities and schools. As part of this initiative, this thesis focuses on the development of a simulation framework that allows for the design and testing of mobile robots across a variety of realistic scenarios. The system is built in ROS 2 and supports both Gazebo Classic and Gazebo Fortress as simulators, with the former intended for lightweight functionalities and the latter for more advanced capabilities and future support. Two robots were integrated into the framework, the Novamob and the Scout, each with different sensor configurations and target use cases. A modular and extensible design approach was adopted to ensure support for multiple robot configurations, sensors, navigation strategies, and both indoor and outdoor environments, facilitating easy customization and future expansion. The results demonstrated that the developed framework stands as an effective simula- tion tool for testing diverse robot configurations, environments, and navigation strategies. The system proved to be adaptable and extensible, providing valuable resources for future users, particularly in educational and research contexts.
- PROJETO DE UM CONVERSOR DE ELETRÓNICA DE POTÊNCIA PARA ENSAIO DE UM CABO SUPERCONDUTORPublication . Lopes, Edgar Esteves Engrácio de Rosado; Vilhena, Nuno; Pina, JoãoNos últimos anos, têm sido realizadas, com carácter experimental, a implementação de cabos supercondutores para se avaliar a possibilidade de reforçar a capacidade das redes elétricas. Tradicionalmente aumentava-se a secção dos condutores para atender à crescente procura por energia, contudo a possibilidade de utilização de cabos supercondutores pode alterar esse paradigma. Apesar do seu potencial, a utilização de cabos supercondutores enfrenta ainda um longo percurso a percorrer, a sua utilização é bastante dispendiosa e levará à necessidade de adaptação das infraestruturas existentes. Neste contexto, as realizações de ensaios a cabos supercondutores tornam-se uma necessidade, de forma a caracterizar o seu comportamento para diferentes regimes de funcionamento. O objeto desta dissertação é desenvolver um conversor de eletrónica de potência, com capacidade de ensaiar cabos supercondutores sob diversos regimes de operação. Os regimes sobre a qual esta dissertação se focou foram os seguintes, transiente e contínuo. Foi realizada uma análise de sistemas de controlo para a aplicação destes regimes de modo a garantir o bom desempenho do conversor a desenvolver. Tomando por base a análise de diversos conversores de eletrónica de potência, definiu- se a topologia que melhor correspondia ao desempenho dos testes que se pretendem realizar aos cabos supercondutores. Pretende-se ainda desenvolver a parametrização da topologia definida para que cumpra os parâmetros. Posteriormente foram realizadas simulações que permitiram avalizar a exequibilidade do projeto. Atendendo à viabilidade do proposto, desenvolveu-se o desenho das placas de circuito impresso, correspondentes ao conversor escolhido.
- Design and Development of the Low Voltage Electrical System for a Formula Student EVPublication . Caldeirinha, Miguel Sobral de Castro; Vilhena, NunoThis dissertation seeks to design and develop an electrical architecture for the low volt- age system of a Formula Student electric vehicle. The objective is to facilitate the car's opera- tion while ensuring its competitiveness and adherence to regulatory standards, so permitting participation in all contests throughout Europe. This topic is highly pertinent to the automotive industry and its competitive aspect, pre- senting a substantial opportunity to develop a system that will allow the team to compete and represent NOVA University throughout Europe, thereby demonstrating the quality and com- petence of NOVA School of Science and Technology (SST) students. This theme brings challenges, as a system with these characteristics has never been de- veloped before by SST students, having to develop everything for the first time and without any previous experience. There are many challenges, of which I highlight the following: I'm not familiar with the operation of an electric vehicle; ii) there is no real product development com- ponent in the course, such as, for example, the design of Printed Circuit Boards (PCBs), which are essential for the correct operation of the vehicle; and finally, iii) depending on an entire team, which, in addition to the regulations, has several design and financial limitations, with an impact on the work I will develop. Taking this into account, I can conclude that the project is enormously challenging and there is real complexity in this project. The selected approach is derived from prior experiences concerning the development of the electrical system for the preceding Combustion Vehicle (CV), developed in previous years by the NOVA Formula Students (NFS), where numerous challenges arose caused by insufficient testing and the considerable complexity of the intended systems. Consequently, the selected strategy will involve creating straightforward systems that are functional and adhere to rules, enabling the team to enhance and expand upon prior concepts in the future. The project begins with a comprehensive examination of the Formula Student Germany (FSG) regulations and an investigation of the system architectures of other teams to determine the essential functional and safety needs. Following this research, a comprehensive low voltage (LV) system architecture was established to facilitate the operation of auxiliary components, reinforce the high voltage (HV) powertrain, and assure rigorous adherence to safety regula- tions. A number of specialized PCBs were developed and further developed to guarantee reli- ability, accessibility, and ease of wiring. The Vehicle Control Unit (VCU) is included among these. Concurrently, critical safety systems including the Shutdown Circuit (SDC), the Brake System Plausibility Device (BSPD), and the Tractive System Active Light (TSAL) were established. The ultimate outcome of this project is a low voltage (LV) system architecture and the corresponding first-generation hardware prototypes. Although the hardware requires further refinement and testing before full deployment, the architecture provides a solid technical foundation for the first generation of NOVA Formula Student electric vehicle. This work establishes a robust baseline for future teams to complete, validate, and improve the system.
- Ultra-low power current and voltage references for in-vivo sensing devicesPublication . Silva, Mariana Sequeira da; Goes, João; O’Hare, DanielWith the continued scaling of technology, devices increasingly rely on smaller batteries, necessitating low-power circuits to ensure sufficient autonomy. This thesis focuses on a biomedical wireless system, in which a stable voltage reference is essential—independent of process, voltage, and temperature (PVT) - to guarantee reliable operation of sensing circuits. This work presents a 3T CMOS voltage reference, in which two MOSFETs in self- cascode configuration generate the proportional-to-temperature component, while a third MOSFET connected in a diode configuration provides the complementary response. This topology overcomes several limitations of 2T references, such as low output voltage, and also offers advantages over conventional bandgap references, including reduced power dissipation, simplified stability, and lower battery requirements. Furthermore, it avoids BJT limitations and parasitic effects in BCD-on-SOI processes that can degrade performance. The circuit incorporates digital startup for both the current reference and the LDO. The current reference supplies the voltage reference, while the LDO regulates battery power. The voltage reference achieves a temperature coefficient (TC) of 300 ppm/°C with the LDO and 30 ppm/°C without it, exhibiting a power supply rejection ratio (PSRR) of -52dB at 100Hz (without LDO) and -53dB (with LDO), while consuming only 50nW. It operates from a 1-1.8 V battery and produces an output of approximately 0.6 V.
- Correspondência de Características Visuais: Comparação entre Métodos Tradicionais e de Deep LearningPublication . Dionísio, Beatriz da Silva; Moutinho, Filipe; Carvalho, JoãoEntre as diferentes aplicações de computer vision, tais como reconhecimento de objetos, reconstrução 3D e localização visual, a correspondência de características visuais é um passo essencial. Esta dissertação tem como objetivo analisar diversos parâmetros, entre eles tempo de execução e robustez, entre métodos tradicionais e algoritmos de deep learning. SIFT, ORB, BRIEF e AKAZE são os métodos tradicionais selecionados para este estudo, pois são conhecidos pelo uso de descritores locais na deteção e correspondência de pontos- chave. Para investigação do lado de deep learning, LoFTR, SuperPoint, SuperGlue e DISK são os algoritmos escolhidos, uma vez que, são métodos que utilizam redes neurais profundas para extrair e associar características visuais. A análise abrange a utilização desses algoritmos para comparar duas imagens, sendo que uma delas contém o objeto de interesse a ser encontrado na outra. Além da precisão na correspondência, são analisados aspetos de desempenho funcional, como a quantidade de descritores produzidos, resistência a variações, duração da execução e uso de memória. Os resultados possibilitam não só verificar as vantagens e limitações de cada algoritmo como também identificar os aspetos onde os métodos tradicionais ainda são competitivos relativamente aos algoritmos de deep learning, permitindo uma escolha mais acertada de acordo com os requisitos do objetivo em mãos. Com este projeto, espera-se contribuir para uma compreensão mais completa e extensa dos diversos algoritmos e das suas capacidades e limitações.
- Target Drone Pose Estimation Towards Drone Capture and LaunchingPublication . Costa, João Henrique Rodrigues Pereira da; Guerreiro, BrunoThe demand for UAVs in modern world applications has seen a steep rise throughout the last decade. Versatility, robustness, and reliability are key factors when it comes to employing these vehicles in tasks such as agriculture, photography, or even military and search-and-rescue operations. This dissertation falls under Project CAPTURE’s scope which intends to address the technological challenges associated with promoting a secure launch and landing of a fixed-wing UAV by leveraging the VTOL capabilities of a rotary-wing drone. The latter, also known as the shuttle drone, incorporates a PX4 autopilot system built upon a ROS2 framework making it possible to implement an onboard MPC, thus fully automating the capture process. The proposed solution aims at removing the current need for cooperation between both UAVs while airborne through the use of an onboard RGB-D camera along with the development of object detection and identification algorithms. Two separate strategies were created: one involving the detection of ArUco markers and their pose in regard to the camera’s POV; another consists on a YOLO-based neural network that was trained according to the fixed-wing drone’s features, thus identifying the target to acquire during the capture manoeuvre. Simulation and tests with the built UAVs reveal promising results with considerable range and high confidence target identification. Still, further development on MPC methods and field testing are required to fully complete the airborne capture manoeuvre.
- Numerical Thermal Analysis of Rocket Engine Nozzle Walls and their Cooling SystemsPublication . Aires, Rafael Alcântara; Albuquerque, Duarte; Vale, João; Gouveia, AlexandreThe thermal loads acting on the walls of a rocket engine nozzle can induce high stresses, which may lead to material plastic deformation/cracks and potential failure. To mitigate these effects, cooling systems are employed to keep wall temperatures within material limits. Among the available systems, regenerative cooling is considered the most efficient for Liquid Rocket Engines (LREs), as it preserves structural integrity and preheats the fuel, thus enhancing overall engine performance. This thesis evaluates the feasibility of performing a fully coupled Computational Fluid Dynamics (CFD) simulation of the nozzle gas flow, coolant, and wall temperature using ANSYS Fluent. This approach contrasts with conventional strategies that decouple the problem. To achieve this, a semi-coupled methodology is followed, first addressing the interaction between the nozzle flow and the wall, and then analyzing the wall with the coolant flow. The case studies in this thesis include grid convergence analyses, evaluations of various Reynolds-Averaged Navier-Stokes (RANS) turbulence models, and comparisons with reference cases from the literature to validate the results in Fluent. In critical wall temperature regions, the results did not deviate by more than 2% compared to other studies. Overall, Fluent accurately captures the thermal behavior in semi-coupled Conjugate Heat Transfer (CHT) simulations, providing a solid basis for extending the methodology to fully coupled analysis of a regenerative cooling system in a rocket. The fully coupled analysis was not performed in this work.
- Anomaly Detection in Fused Deposition Modeling with Deep Learning and Computer VisionPublication . Costa, Hugo do Carmo Lopes dos Santos; Moutinho, Filipe; Bragadesto, DuarteThe introduction of Industry 4.0 concepts has fueled the advancement of modern technologies such as additive manufacturing and artificial intelligence. Interchangeably referred to 3D printing, additive manufacturing has gained widespread popularity not only for facilitating rapid prototyping, essential for product development, but also for its ability to manufacture localized customized parts without the need for special tooling. An additional advantage stems from the low initial investment cost, which makes the technology accessible. This is the case with Fused Deposition Modeling 3D printers. Traditionally, the quality assessment and monitoring of active 3D printing jobs have required significant human intervention, which can be imprecise. Anomalies might be overlooked by the human eye, and in some cases, this approach can be destructive to the printed object, leading to a reprint job and increased material wastage. The present document discloses available state-of-the-art systems that confer automatic anomaly detection on ongoing 3D printing jobs through the use of deep learning networks and visual data. Furthermore, an enhancement to a state-of-the-art system which operates using this type of deep learning framework is proposed. It is theorized that the proposed modifications will improve the system’s performance and yield favorable results. Additionally, to verify the developed system reproducibility and contribute to the existing dataset of the stateof- the-art system, a dataset containing material deposition images will be developed.